fork()

Updated: April 19, 2023

Create a new process

Synopsis:

#include <sys/types.h>
#include <unistd.h>

pid_t fork( void );

This function is declared in <process.h>, which <unistd.h> includes.

Library:

libc

Use the -l c option to qcc to link against this library. This library is usually included automatically.

Description:

The fork() function creates a new process. The new process (child process) is an exact copy of the calling process (parent process), except for the following:
  • The child process has a unique process ID.
  • The child process has a different parent process ID (which is the process ID of the calling process).
  • The child process has its own copy of the parent's file descriptors. Each of the child's file descriptors refers to the same open file description with the corresponding file descriptor of the parent.
  • The child process has its own copy of the parent's open directory streams.
  • The child process's values of tms_utime, tms_stime, tms_cutime, and tms_cstime are set to zero.
  • The child process has a single thread, regardless of the number that the parent has.
  • File locks previously set by the parent aren't inherited by the child.
  • The child process doesn't inherit any timers, QNX Neutrino message-passing objects, side-channel connection IDs (coids), or channels (chids).
  • (QNX Neutrino 7.1 or later) The child process doesn't inherit I/O privileges.
  • The set of signals pending for the child process is initialized to the empty set.
  • No memory locks set by the parent process are inherited by the child. For more information, see Locking memory in the Process Manager chapter of the System Architecture guide.
  • No event notification requests, such as those requested with ionotify(), mq_notify(), procmgr_event_notify(), procmgr_value_notify(), or SyncMutexEvent(), are inherited by the child.
  • No interrupt handlers or events attached to interrupt sources through InterruptAttach(), InterruptAttachArray(), and InterruptAttachEvent() are inherited by the child.
Note that:
  • A named semaphore (sem_open()) increments and decrements the same count in the parent and child, but an unnamed semaphore (sem_init()) increments and decrements a different count (with the same initial value) in the two processes, unless that semaphore is in a shared memory object.
  • The child inherits any message queue descriptors from the parent, whether you're using the traditional (mqueue) or alternative (mq) implementation of message queues.
  • The child inherits any memory mappings that were created in the parent, except for those that (in QNX Neutrino 7.1 or later) have the MAP_NOINHERIT flag set (see mmap()).

    MAP_PRIVATE mappings inherited from the parent are also MAP_PRIVATE mappings in the child, and any modifications to the data in these mappings made by the parent prior to calling fork() are visible to the child. Any modifications to the data in MAP_PRIVATE mappings made by the parent or child after a fork() are visible only to the parent or child, respectively.

  • The parent's ability configuration (see procmgr_ability()) is copied to the child verbatim, regardless of the PROCMGR_AOP_INHERIT_NO status of each of the abilities.
  • If the parent is a resource manager, then the child can't be a resource manager until after an exec().
  • If the parent has used any of the aio_*() functions, then the child can't use any of the aio_*() functions until after an exec().
  • If the child will link a library at load time or at run time (i.e., with dlopen()), the child must be given the PROCMGR_AID_PROT_EXEC and PROCMGR_AID_MAP_FIXED abilities. For more information, see procmgr_ability().

In order to successfully call this function, your process must have the PROCMGR_AID_FORK ability enabled. For more information, see procmgr_ability().

You can use pthread_atfork() to register fork handler functions to be called before or after the fork occurs.

Note:
  • POSIX requires that the child process use only functions that are async-signal-safe until it calls one of the exec*() functions. If your program uses mutexes, you might need to use a forksafe mutex or register a pthread_atfork() handler that locks all the mutexes before you fork.
  • A thread must not hold a forksafe_mutex_t when it invokes fork(), or else a deadlock will occur. The same is true of a pthread_mutex_t that's protected by an at-fork handler.

For more information, see Using fork() in a multithreaded process in the “Processes and Threads” chapter of Getting Started with QNX Neutrino.

Returns:

A value of zero to the child process, and the process ID of the child process to the parent process. Both processes continue to execute from the fork() function. If an error occurs, fork() returns -1 to the parent and sets errno.

Errors:

EAGAIN
Insufficient resources are available to create the child process. For example, you might have exceeded the maximum number of processes permitted; see the RLIMIT_NPROC resource for getrlimit().
EBADF
A problem occurred when fork() was duplicating a file descriptor. For example, another thread might have opened or closed a file descriptor while the fork() was occurring. You can add synchronization around the operations that involve file descriptors, or try calling fork() again.
ENOMEM
The process requires more memory than the system is able to supply.
ENOSYS
The fork() function isn't implemented for this memory protection model.
EPERM
One of the following occurred:
  • The calling process doesn't have the required permission; see procmgr_ability().
  • (QNX Neutrino 7.0.1 or later) The calling process's priority is above the permitted range.

Examples:

This program demonstrates forking. You run it like this:

fork_example program [arguments]

The child runs program (passing it the arguments, if any), and the parent waits for the child to exit.

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/wait.h>

int main( int argc, char **argv )
{
    pid_t pid;
    pid_t wpid;
    int   status;

    if (argc == 1) {
        printf ("Usage: %s <program> [<arguments>]\n", argv[0]);
        return EXIT_FAILURE;
    }

    if( ( pid = fork() ) == -1 ) {
        perror( "fork" );
        return EXIT_FAILURE;
    }

    if( pid == 0 ) {
        printf ("Child: My pid is %d.\n", getpid() );

        /* Use exec to become the specified program. */
        execvp( argv[1], argv+1 );

        /* This can happen only if execvp() fails; print a message, and then exit. */
        perror( argv[1] );
        return EXIT_FAILURE;

    } else {
        printf ("Parent: My pid is %d.\n", getpid() );

        /* Wait for the child to finish. */
        do {
            wpid = waitpid( pid, &status, 0 );
        } while( WIFEXITED( status ) == 0 );

        printf ("Parent: Child %d has exited with status %d; we're done!\n",
                wpid, WEXITSTATUS( status ));
        return WEXITSTATUS( status );
    }
}

Classification:

POSIX 1003.1

Safety:  
Cancellation point No
Interrupt handler No
Signal handler Yes
Thread Read the Caveats

Caveats

It's possible to call fork() from a multithreaded process, but it can be very difficult to do so safely, so we recommend that you call it only from single-threaded processes. For more information, see Using fork() in a multithreaded process in the “Processes and Threads” chapter of Getting Started with QNX Neutrino.